Over the last several years, the Bureau of Labor Statistics has consolidated several of its compensation surveys into the National Compensation Survey (NCS). Combining the separate surveys into one integrated program provides greater efficiency in data collection and processing, and also provides greater flexibility in the calculation of compensation measures. Several new statistics are planned for publication from the NCS, which should give data users a more complete understanding of the compensation package a U.S. worker typically receives.
Data are available from the first collection period of the NCS. It is now possible to calculate trial estimates for the proposed statistics. This article describes the integration of the separate surveys into the NCS and presents trial statistics for health insurance. (1) The calculation of the trial statistics is just the first step in the development of the new measures. It will take some time to incorporate their calculation into the production process for regular release. Nonetheless, the statistics in this article are examples of the types of measures BLS hopes to publish in the future.
Previous compensation surveys
The NCS combined the Employment Cost Index (ECI) Survey, the Employee Benefits Survey (EBS), and the Occupational Compensation Survey (OCS) Program. The ECI has been the most prominent of the compensation surveys. It shows employers' costs for wage and nonwage compensation relative to a base period using a Laspeyres formula. (2) Nonwage compensation covers an extensive list of employee benefits, including health insurance, pensions, paid leave, and legally-required benefits. BLS began reporting the ECI for benefits in 1981. From that time until 1995, growth in benefit compensation consistently outpaced growth in wage compensation. Then, during the second half of the 1990s, the trend reversed and benefits grew at a lower rate than wages. Since 2000, benefit costs have accelerated, growing more quickly than wages once again. (3)
As a measure of total compensation for U.S. workers, the ECI is the chief indicator of compensation inflation for the U.S. labor market. However, it does not provide any detail on the benefit packages workers receive. As a step to providing such detail, BLS introduced the Employer Costs for Employee Compensation (ECEC) in 1987. (4) The ECEC is based on the ECI data for the current period. It reports average compensation for wages and salaries and for 19 categories of benefits. In December 2003, the ECEC for total compensation was $22.92 per hour worked for workers in private industry. Wages and salaries made up 71.9 percent of compensation, with an average of $16.49. Benefits made up the remaining 28.1 percent, with an average of $6.43. Of particular interest among the individual benefits, the cost per hour worked for health insurance averaged $1.50 in December 2003, or 23 percent of total benefit costs.
Although the ECEC provides a good summary of the relative cost for the various pieces of the benefit package a worker typically receives, it does not describe the characteristics of those benefits. Detailed characteristics of plans have historically been the purview of the Employee Benefits Survey (EBS). The EBS reported the proportion of employees who participate in benefit plans with a particular provision or characteristic. The provisions ranged from the very broad, such as whether a health plan included dental coverage, to the very specific, such as whether a medical plan's limit to inpatient alcohol detoxification coverage was measured in days rather than dollars.
The EBS provided rich detail about benefit plans, but its drawback was the lag time between data collection and the published statistics. In contrast, the ECI is reported in the month after its reference period, and the ECEC is reported shortly thereafter, so their information is quite timely. Moreover, statistics from the ECI and the EBS programs did not connect employers' costs for the benefits with the provisions of the benefit plans. The integrated NCS allows both the timelier release of some of the data in plan provisions statistics and the publication of these data that link the costs of plans to their provisions.
The integrated NCS
Some integration of the ECI and EBS predates the introduction of the National Compensation Survey. The EBS sample was made up of establishments in the ECI sample as of the August prior to the EBS reference year. Establishments also needed to satisfy the size and ownership restrictions for the year, as the EBS reported statistics for medium and large private establishments in odd-numbered years, and for small private establishments and State and local government establishments in even-numbered years. (5) After the ECI determined the composition of the EBS sample, however, their data collection was separate, which precluded combining the cost information with the provision information from the surveys.
The NCS completes the integration of the ECI and EBS. Cost, provision, and participation data will be collected by benefit plan, which allows the costs of plans to be linked with their provisions. The cost data will be updated every 3 months, as required by the ECI, for the approximate 5 years the establishment is scheduled to remain in the NCS survey. Participation rates in the benefit plans, along with some broad provisions of the plans, will be updated annually to keep the information current. The more detailed characteristics of plans take longer for BLS to collect, compile, and verify; they will be collected exclusively when the establishment initially enters the sample.
Trial statistics for health insurance
Trial statistics for health insurance were calculated based on the integrated NCS data collection. The sample is restricted to private establishments. The reference month for most of the data is June 2003; almost all of the remaining data refer to either May or July 2003. The sample of establishments matches the sample used for the NCS statistics in "Employee Benefits in Private Industry, 2003" News Release USDL: 03-489.
There are four types of trial statistics: cost per employee, access rates, participation rates, and cost per participant. The ECEC has always measured the employer costs per employee, so the cost-per-employee statistics follow the formula used for the ECEC. (6) The EBS measured participation rates in benefit plans with particular provisions, so the participation-rate statistics follow the formula used historically by the EBS. (7) Cost-per-participant statistics are new under the NCS; they equal the sum of costs among benefit plans with a particular provision divided by the total number of participants in plans with the provision.
Access-rate statistics are also new under the NCS. To understand their calculation, it is important to become familiar with the sampling scheme used by the NCS. Sampling proceeds as follows:
* Establishments are selected for the sample.
* Within each selected establishment, a small number of individual employees are selected from an employee list.
* For each selected worker, the establishment is asked to define the worker's job according to the establishment's most detailed classification system.
* Data then are collected for all workers who hold the job.
The NCS does not use individual workers as its unit of observation because the ECI needs the unit to remain intact if a worker leaves the establishment or switches to another job.
In the calculation of the access-rate statistics, all workers in the occupation are assumed to have access to a benefit plan if the establishment offers a plan to at least one of the workers in the occupation. The access rate then equals the number of employees with access to a plan divided by the total number of employees. (See box.)
Shown below is the first set of trial statistics on health insurance for workers in private industries in 2003. (8) The bold type highlights statistics that are new under the NCS. The statistics are calculated using estimation methods that sometimes differ from the methods used for statistics published from the 2003 NCS. The units for the cost statistics are dollars per hour worked by the employee.
As noted, the cost per employee for health insurance is $1.39 per hour worked. This statistic corresponds conceptually to the ECEC for health insurance, which is published every 3 months. (9) Health insurance plans are then divided into three types: medical, stand-alone dental, and stand-alone vision. Medical plans that also have dental or vision coverage are included with medical to eliminate overlap among the categories. Of the employers' average cost of $1.39 per hour, about 14 cents are for nonmedical health plans.
